ARTICLE IN PRESS

Journal of Behavior Therapy

and Experimental Psychiatry 37 (2006) 206–212

0005-7916/$ -

doi:10.1016/j

�CorrespoE-mail ad

www.elsevier.com/locate/jbtep

Metacognitive therapy for generalized anxietydisorder: An open trial

Adrian Wellsa,�, Paul Kingb

aUniversity of Manchester, Academic Division of Clinical Psychology, Rawnsley Building, MRI,

Manchester M13 9WL, UKbManchester Mental Health Partnership, Manchester, UK

Received 4 November 2004; received in revised form 7 July 2005; accepted 18 July 2005

Abstract

Generalized anxiety disorder (GAD) responds only modestly to existing cognitive-

behavioural treatments. This study investigated a new treatment based on an empirically

supported metacognitive model [Wells, (1995). Metacognition and worry: A cognitive model

of generalized anxiety disorder. Behavioural and Cognitive Psychotherapy, 23, 301–320; Wells,

(1997). Cognitive therapy of anxiety disorders: A practice manual and conceptual guide.

Chichester, UK: Wiley]. Ten consecutive patients fulfilling DSM-IV criteria for GAD were

assessed before and after metacognitive therapy, and at 6, and 12-month follow-up. Patients

were significantly improved at post-treatment, with large improvements in worry, anxiety, and

depression (ESs ranging from 1.04–2.78). In all but one case these were lasting changes.

Recovery rates were 87.5% at post treatment and 75% at 6 and 12 months. The treatment

appears promising and controlled evaluation is clearly indicated.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Generalized anxiety disorder; Cognitive therapy; Worry; Metacognition; Trait anxiety

see front matter r 2005 Elsevier Ltd. All rights reserved.

.jbtep.2005.07.002

nding author. Tel.: +0044 161 276 5387.

dress: adrian.wells@man.ac.uk (A. Wells).

ARTICLE IN PRESS

A. Wells, P. King / J. Behav. Ther. & Exp. Psychiat. 37 (2006) 206–212 207

1. Introduction

Generalized anxiety disorder (GAD) appears moderately responsive to cognitive-behavioural treatments (e.g. Durham & Allan, 1993). In a reanalysis of data from sixCBT outcome studies, Fisher and Durham (1999) reported a recovery rate across alltreatments of 40% overall based on trait-anxiety scores (Speilberger, Gorsuch,Lushene, Vagg, & Jacobs, 1983). Two treatments, applied relaxation (AR) andindividual cognitive behaviour therapy (CBT), did best with recovery rates at posttreatment of 17–59% for AR and 26–71% for CBT. At 6-month follow-up oneparticular study (Borkovec & Costello, 1993) obtained a recovery rate for ARof 81%.

In two more recent studies, AR appeared less effective (Arntz, 2003; Ost &Breitholtz, 2000). Ost and Brietholtz obtained small improvements in trait anxietyfollowing AR. Arntz (2003) compared cognitive therapy with applied relaxation. Atpost treatment he reported that 35% of cognitive therapy patients and 44.4% ofapplied relaxation patients were recovered. At 6-month follow-up this had increasedto 55% of cognitive therapy patients and 53.3% of applied relaxation patients on thebasis of the trait-anxiety scale.

These data show that the outcomes for AR and CBT show considerablevariability, and there is a need for more effective treatments. Recent attempts toimprove treatment have combined these treatment elements, and increased theamount of therapy delivered (e.g. Borkovec, Newman, Pincus, & Lytle, 2002;Durham et al., 2004). However, so far treatment outcomes have not improved.

Progress might be made by basing treatment on a model of the mechanisms andfactors underlying pathological worry, the hallmark of this disorder. The presentstudy reports an initial evaluation of a new form of cognitive therapy (metacognitivetherapy (MCT): Wells, 1995, 1997) that is based on a specific model of GAD.Furthermore, it aims to assess the impact of the treatment on multiple dimensions ofworry.

The metacognitive model (Wells, 1995, 1997) asserts that individuals with GAD,like most people, hold positive beliefs about worrying as an effective means ofdealing with threat. However, worry is used as an inflexible means of coping, and thisbecomes a problem when negative beliefs concerning the uncontrollability and thedangers of worrying develop, leading to unhelpful control strategies.

In this model two broad subtypes of worry are distinguished called type 1 and type2 worry. Type 1 refers to worry about external events and physical symptoms, andcan be distinguished from type 2, which concerns negative appraisals of worrying.Essentially type 2 worry is worry about worrying. In the model worrying is used as ameans of coping with threat. It persists until the individual achieves an internal/external signal that signifies that it is safe to stop worrying or until the person isdistracted from the activity. During the development GAD negative appraisals ofworrying and associated negative beliefs about worry develop. Two domains ofnegative belief/appraisals are important and concern (1) the uncontrollability ofworrying, and (2) its dangerous consequences for physical, psychological, and socialfunctioning. When negative metacognitions of this kind develop, the person

ARTICLE IN PRESS

A. Wells, P. King / J. Behav. Ther. & Exp. Psychiat. 37 (2006) 206–212208

experiences an elevation in distress and worry. The co-existence of positive andnegative beliefs about worrying lead to unhelpful vacillation in attempts to avoid andengage in worry, and the use of unhelpful mental regulation strategies such asreassurance seeking and thought suppression. Such strategies when they aresuccessful prevent the person from discovering that worrying does not lead tocatastrophe. Some strategies do not work and reinforce beliefs in loss of control. Forexample, attempting to suppress thoughts that trigger worry can backfire andincrease preoccupation with these thoughts. Strategies such as seeking reassurancedo not allow the person to unambiguously discover that worrying can be controlledby the self. It follows from this model that successful treatment of GAD should focuson modifying several metacognitive factors, including counterproductive thoughtcontrol strategies, erroneous beliefs about the uncontrollability of worry, negativebeliefs about the danger of worrying, and positive beliefs that support the over-reliance on worrying as a coping strategy.

2. Method

2.1. Participants and design

Patients were drawn from consecutive referrals made by general practitioners andpsychiatrists to two NHS clinical psychology departments. Diagnosis wasestablished using the structured clinical interview for DSM-IV. Patients wereincluded if GAD was their primary problem. Patients who had received previouscognitive-behavioural treatment for GAD were excluded. Ten patients wererecruited, six of these were female and four were male, and the ages of subjectsranged from 25 to 76 years. None of the patients were currently taking psychotropicmedication (two patients had previously taken benzodiazepines, and four patientshad previously used medications that they were unable to specify). One of thepatients was a student and four were professional workers, employed in white-collarwork. The remaining five patients were retired or unemployed.

The duration of GAD ranged from 2 to 60 years. Fifty per cent of patients had asingle diagnosis of GAD, and 50% had additional diagnoses. Three patients (30%)met criteria for additional major depressive disorder, one patient (10%) met criteriafor social phobia, and one patient (10%) had depression not otherwise specified andsocial phobia. We did not screen for axis II disorders. Once the presence of DSM-IVGAD was established using the SCID, patients were asked: ‘‘If treatment wassuccessful in alleviating your worry and associated anxiety would you requiretreatment for any other problems?’’ If patients responded ‘‘No’’ they were includedin the study. Only one patient was excluded on this basis.

2.2. Measures

The following outcome measures were completed at the beginning and end ofbaseline, end of treatment and at follow-up: Beck anxiety inventory (BAI; Beck,

ARTICLE IN PRESS

A. Wells, P. King / J. Behav. Ther. & Exp. Psychiat. 37 (2006) 206–212 209

Epstein, Brown, & Steer, 1988), Beck depression inventory BDI (Beck, Ward,Mendelson, Mock, & Erbaugh, 1961), trait-anxiety subscale of the state trait-anxietyinventory (Speilberger et al., 1983), and anxious thoughts inventory (AnTI: Wells,1994, 2000). The AnTI is a multidimensional measure of proneness to subtypes ofworry. It has three subscales: social worry, health worry, and meta-worry (worryabout worry).

Patients received individual weekly sessions of metacognitive therapy (MCT) of45–60min duration. Homework was a component of treatment, and treatmentfollowed the manual by Wells (1997, pp. 200–235).

3. Results

We had initially aimed to offer 4–12 treatment sessions, since this was the rangeeffective in previous work. However, the minimum number of sessions received byone patient was three. This patient reported significant improvement in symptomsafter the third session, and she requested treatment termination prior to attendingsession 4. Therefore, the range of treatment sessions offered was 3–12. Two patientsreturned incomplete trait-anxiety measures at pre-treatment and so pre-treatmentdata on this measure are missing for these two cases; consequently all analysis oftrait-anxiety scores is based on the remaining eight cases.

Examination of pre-treatment scores shows that the patients in this trialhad marginally lower mean trait-anxiety scores than those reported by Borkovecet al. (2002). However, trait-anxiety levels were similar to the levels reportedby Ost and Breitholtz (2000) and Arntz (2003) in their applied relaxation treatmentconditions, and by Borkovec and Costello (1993) in CBT. Furthermore, pre-treatment worry scores assessed by the AnTI in the present study were similarto those reported by Durham et al. (2004) in their treatment study. Thus, thecurrent sample appears to consist of patients comparable with patients treated inother trials.

All patients improved during the course of treatment. Table 1 presents meanscores on each outcome measure at pre-treatment, post-treatment, and at follow-up.Paired samples t-tests showed significant improvements in all measures at post-treatment, and gains remained significant at follow-up assessments.

Post-treatment effect sizes (Cohen’s d, 1977: M1–M2/sd M1) were verylarge: BAI ¼ 1.82, trait-anxiety ¼ 2.78, meta-worry ¼ 1.47,social worry ¼ 1.13,health worry ¼ 1.12, BDI ¼ 1.41. At 6 and 12-month follow-up the Es were:BAI ¼ 1.63, 1.69, trait-anxiety ¼ 2.46, 2.58, meta-worry ¼ 1.30, 1.61, socialworry ¼ 1.10, 1.11, health worry ¼ 1.12, 1.04, BDI ¼ 1.22, 1.28.

Of particular theoretical interest, the results appear to show that treatmentdirected specifically at modifying metacognitions leads to substantial decreases in alldimensions of worry (social and health), as well as worry about worry (meta-worry).

Treatment also seems to have significantly impacted on the more somaticmanifestations of anxiety as indexed by scores on the BAI. Similar effects wereobserved for depressed mood (BDI).

ARTICLE IN PRESS

Table 1

Means, standard deviations and t statistics for outcome measures at pre-treatment (Pre), post-treatment

(Post), and follow-up assessments

Measure Pre Post Follow-up t-values

6-month 12-month Pre/Post Pre/6m Pre/12m

M (SD) M (SD) M (SD) M (SD) t p t p t p

BAI 21.00 (9.67) 3.40 (4.12) 5.20 (5.16) 4.70 (5.29) 6.14 o0.0005 4.30 0.002 4.38 0.002

AnTI-S 19.60 (6.20) 12.60 (4.33) 12.80 (3.82) 12.70 (4.03) 5.22 0.001 5.42 o0.0005 4.08 0.003

AnTI-H 12.10 (3.57) 8.10 (1.79) 8.10 (1.91) 8.40 (2.80) 3.83 0.004 4.00 0.003 2.68 0.03

AnTI-M 18.10 (5.09) 10.60 (3.10) 11.50 (3.21) 9.90 (2.92) 6.83 o0.0005 4.94 0.001 4.92 0.001

BDI 13.90 (7.72) 3.00 (3.59) 4.50 (4.90) 4.00 (5.23) 6.87 o0.0005 4.09 0.003 4.31 0.002

STAI-T 52.50 (7.07) 32.88 (6.64) 35.13 (12.36) 34.29 (14.38) 5.37 0.001 3.94 0.006 3.42 0.01

Note: BAI ¼ Beck anxiety inventory; AnTI ¼ anxious thoughts inventory; (S ¼ Social worry, H ¼ health

worry; M ¼ meta-worry), BDI ¼ Beck depression inventory; STAI-T ¼ trait anxiety.

A. Wells, P. King / J. Behav. Ther. & Exp. Psychiat. 37 (2006) 206–212210

The mean improvement in trait-anxiety at post-treatment was 19.63. Fisher andDurham (1999) used Jacobsen criteria for defining clinically significant changeparameters in trait-anxiety. Using these criteria it is possible to classify patients asworse, improved, or recovered. From the eight patients who returned unspoiledSTAI-T scales, 87.5% of patients met criteria for recovery at post-treatment and allof them met the criterion of clinically significant improvement. At 6-month follow-up one patient’s STAI-T score had returned to pre-treatment levels whilst the scoresof each of the other patients retained the recovery status. Thus, 75% of patientsremained recovered at 6 and 12-month follow-up. These results compare well againstthe six CBT trials analyzed by Fisher and Durham (1999) in which a recovery rate of40% was found for the sample overall at 6-month follow-up, and for the bestexamples involving individual applied relaxation or cognitive-behaviour therapy thiswas 81% and 65%, respectively. More recent studies have produced lower figures(Arntz, 2003; Durham et al., 2004).

4. Discussion

All patients were improved on self-report measures at post-treatment, and effectswere maintained at follow-up in all but one case. The degree of improvement acrossmeasures suggests that treatment was highly effective. The range of sessions offeredwas 3–12 with a mean of 7.4, suggesting that MCT is economical to use.

Effect sizes were very large at post-treatment and at follow-up. The effect sizeswere larger than those typically obtained in evaluations of treatment for GAD.Similarly, recovery rates appeared much higher at post-treatment (87.5%) thanobtained in each of the six studies analyzed by Fisher and Durham (1999), orreported in more recent studies (Arntz, 2003; Durham et al., 2004). Recovery rateswere also much higher at follow-up (75%), with the exception of one study reporting

ARTICLE IN PRESS

A. Wells, P. King / J. Behav. Ther. & Exp. Psychiat. 37 (2006) 206–212 211

recovery rates of 81% at 6-month follow-up (59% post-treatment) after appliedrelaxation (Borkovec & Costello, 1993).

Examination of the effects of treatment on different dimensions of worry suggeststhat a specific focus on metacognitions has a general impact across components ofworry. In particular, social and health concerns decreased significantly. This isinteresting because at no stage in treatment was the content of worry, outside of thedomain of metacognition, the focus of modification. Such an approach has theadvantage of avoiding the difficulty encountered by therapists of having torepeatedly pursue the patient’s latest worry.

There are limitations with the present study that must be borne in mind. Theresults are based on a relatively small number of cases and so caution should beused in interpreting the data. Patients were asked if they might require treatmentfor additional problems as a treatment exclusion criterion. Although only onepatient was excluded on this basis, this could produce a biased sample of patientswho do not have secondary problems. However, this is not likely to be a threat toexternal validity in the present case since half of the sample had additional axis Idiagnoses. Furthermore, patients appeared similar in trait-anxiety, worry levels, andmedication status to patients treated in several other studies reported in theliterature.

A reliance strictly on self-report measures of treatment outcome is a limitation forinterpreting the present treatment effects. The absence of a no-treatment controlgroup means spontaneous fluctuation in symptoms cannot be ruled out; however,analysis of wait-list patients from other trials suggests very little evidence ofspontaneous recovery (5%) in untreated patients (Fisher & Durham, 1999).

The results of this preliminary study appear encouraging and support thecontinued evaluation of MCT, which should now be compared with activetreatments such as applied relaxation or standard cognitive behaviour therapy.

References

Arntz, A. (2003). Cognitive therapy versus applied relaxation as treatment of generalized anxiety disorder.

Behaviour Research and Therapy, 41, 633–646.

Beck, A. T., Epstein, N., Brown, & Steer (1988). An inventory for measuring clinical anxiety: psychometric

properties. Journal of Consulting and Clinical Psychology, 56, 893–897.

Beck, A. T., Ward, C. H., Mendelson, M., Mock, & Erbaugh (1961). An inventory for measuring

depression. Archives of General Psychiatry, 4, 561–571.

Borkovec, T. D., & Costello, E. (1993). Efficacy of applied relaxation and cognitive behavioral therapy in

the treatment of generalized anxiety disorder. Journal of Consulting and Clinical Psychology, 61,

611–619.

Borkovec, T. D., Newman, M. G., Pincus, A. L., & Lytle, R. (2002). A component analysis of cognitive-

behavioral therapy for generalized anxiety disorder and the role of interpersonal problems. Journal of

Consulting and Clinical Psychology, 70, 288–298.

Cohen, J. (1977). Statistical power analysis for the behavioral sciences (Revised edition). New York:

Academic Press.

Durham, R. C., & Allan, T. (1993). Psychological treatment of generalized anxiety disorder: A review of

the clinical significance of results in outcome studies since 1980. British Journal of Psychiatry, 163,

19–26.

ARTICLE IN PRESS

A. Wells, P. King / J. Behav. Ther. & Exp. Psychiat. 37 (2006) 206–212212

Durham, R. C., Fisher, P. L., Dow, M. G. T., Sharp, D., Power, K. G., Swan, J. S., et al. (2004). Cognitive

behaviour therapy for good prognosis and poor prognosis generalized anxiety disorder: A clinical

effectiveness study. Clinical Psychology and Psychotherapy, 11, 145–157.

Fisher, P. L., & Durham, R. C. (1999). Recovery rates in generalized anxiety disorder following

psychological therapy: An analysis of clinically significant change in the STAI-T across outcome

studies since 1990. Psychological Medicine, 29, 1425–1434.

Ost, L. G., & Breitholtz, E. (2000). Applied relaxation vs. cognitive therapy in the treatment of generalized

anxiety disorder. Behaviour Research and Therapy, 38, 777–790.

Speilberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, & Jacobs (1983). Manual for the state-trait anxiety

inventory. Palo Alto, CA: Consulting Psychology Press.

Wells, A. (1994). A multidimensional measure of worry: Development and preliminary validation of the

anxious thoughts inventory. Anxiety Stress and Coping, 6, 289–299.

Wells, A. (1995). Meta-cognition and worry: A cognitive model of generalised anxiety disorder.

Behavioural and Cognitive Psychotherapy, 23, 301–320.

Wells, A. (1997). Cognitive therapy of anxiety disorders: A practice manual and conceptual guide.

Chichester, UK: Wiley.

Wells, A. (2000). Emotional disorders and metacognition: Innovative cognitive therapy. Chichester, UK:

Wiley.